See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/291598605

Novel Testing of a Biological Safety Cabinet using PCR

Article · December 2010

DOI: 10.1177/153567601001500405

CITATION READS
1 116

4 authors, including:

Robert A Colvin
Ohio University
61 PUBLICATIONS   1,919 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Robert A Colvin on 16 May 2016.

The user has requested enhancement of the downloaded file.

Articles
Novel Testing of a Biological Safety Cabinet Using PCR
Charles P. Fontaine1, Timothy J. Ryan2, Peter W. Coschigano2, and Robert A. Colvin2
1Diagnostic Hybrids, Inc., Athens, Ohio and 2Ohio University, Athens, Ohio
Abstract
The biological safety cabinet is commonly em-
ployed to protect both user and product in an increas-
ing number of biotechnology applications. It is recom-
mended that cabinets be fully certified annually, but
given the requirements of good laboratory manufactur-
ing practices, more frequent, less disruptive user
checks of equipment may be desirable. Furthermore,
annual field certifications only determine that cabinets
meet basic design and performance parameters and
cannot reveal all issues unique to a particular cabinet.
This study describes the development of a novel in-
laboratory test method using common biotechnology
tools that is capable of assessing bioaerosols contain-
ment as well as cabinet-specific lateral motion of
bioaerosols potentially responsible for product cross-
contamination. Genetically engineered viable bacteria
releases, followed by colony recovery and gene amplifi-
cation by polymerase chain reaction, constitute the es-
sence of the new approach. Specifically, a nebulizer
was used to release a bioaerosol within a cabinet. Sin-
gle-stage bioaerosol samplers were simultaneously
used to sample the challenge bioaerosol containing
bacteria engineered with a plasmid conferring Kana-
mycin resistance. Recovered challenge bacteria were
grown on media containing the antibiotic, and the pres-
ence of a DNA fragment in the bacterial plasmid was
confirmed. Results demonstrated that there was detect-
able lateral motion of bioaerosols within the biological
safety cabinet tested. The importance of such lateral
motion to product cross-contamination is discussed, as
are the applications for this novel testing method. It is
concluded that the new technique represents an en-
hanced level of safety cabinet field testing ability avail-
able to the end user.
Keywords
PCR, laminar flow, NSF 49 testing, safety cabinet,
HEPA, and efficiency
Introduction
Biological Safety Cabinets (BSCs) are employed in
a multitude of clinical, research, and industrial applica-
tions, including blood, bioaerosol, and wastewater sam-
ple analysis, immunology, cell culture, and microbiology.
In such work, reproducibility and the prevention of con-
tamination are paramount to useable outcomes. The
186 www.absa.org Applied Biosafety
BSC is designed to achieve these goals by filtering and
controlling the flow of air over objects contained within
(U.S. Department of Health and Human Services 2007a,
2007b).
BSCs were developed to protect contained materials
and samples (“product”) from contamination arising out-
side of the cabinet, from cross-contamination, and to
protect the user from materials within (Stimpfel & Ger-
shey, 1991). This is achieved primarily through 3 con-
trols, the High Efficiency Particulate Air (HEPA) supply/
exhaust filters, unidirectional flow of the air internally,
and inward air flow at the front entrance to the cabinet.
The prevention of contamination between samples is
intended to result from the unidirectional downward flow
of the air within the work area (Heinsohn et al., 1995).
The Class II, Type A safety cabinet typified by this study
is frequently employed and is detailed in Figure 1. In this
type of cabinet air that has been HEPA filtered is pushed
through an overhead diffuser into the cabinet work-
space. As the air hits the work space it is split and col-
lected by air returns at both the front and rear of the
cabinet work area. At the front air intake, contaminated
air from the room also enters and is pulled down in to
the plenum of the cabinet. Air is then HEPA filtered be-
fore entering the work space via the overhead filter,
thereby protecting cabinet contents from external con-
taminants. This intake of air at the front portal also func-
tions to provide personnel protection by preventing cabi-
net air release outward through this air curtain. Class II,
Type A cabinets typically vent the exhaust air directly into
the room through a second HEPA filter on top of the unit.
In this manner the indoor environment is protected from
any aerosols released by procedures in the cabinet.
The Class II, Type A is manufactured by several com-
panies and is anecdotally considered the most common-
ly encountered BSC. Regardless of the make all similar
class/type cabinets operate under the same design
parameters. These are prescribed by NSF International
in NSF/ANSI Standard 49 (NSF, 2008). In this type of
cabinet approximately 70% of the air contained within
has been recycled through the supply filter and about
30% comes in through the intake at the front. The speci-
fications include the front sash height and opening size,
the amount of air recirculated in the cabinet, and the
intake velocity at the front, among other requirements
(Heinsohn et al., 1995). NSF 49 covers not only con-
struction and design specifications, but also the verifica-
tion methods for examining user and product contamina-
tion protection. Neither protection of the user nor sam-
ple cross-contamination is quantitatively assessed by
Vol. 15, No. 4, 2010
 Articles

Figure 1
Class II Type A Biological Safety Cabinet
A—Contaminated room air. B—Sash, glass or other suitable material. C—HEPA filtered exhaust. D—Sealed return
air plenum. E—HEPA filter for work area airflow. F—Fan. (U.S. Department of Health and Human Services, 2007a).

the current NSF 49 field certification protocol. NSF 49
field tests include air flow velocity and HEPA filter integri-
ty but a field method to determine actual bioaerosol mo-
bility would be of value.
Different aerosol types are generated by many meth-
ods during routine sample and culture handling. Pipet-
ting is one of the common sources of smaller aerosols,
(Gerhardt, 1994) and it has been shown that up to
15,000 particles can be released when the residual con-
tents in a pipette are forcibly expelled (Heinsohn et al.,
1995). Aerosols are created when a pipette is used to
mix a liquid by “bubbling” that liquid, as bursting bub-
bles are known to release droplets into the air. The sim-
ple but routine action of opening containers also has the
potential to generate aerosols. These droplets are gener-
ated when some of the fluid within the container has
come in contact with the cap or plug. As the container is
opened the surface tension holding the fluid between
the container and cap by is broken, releasing droplets
(Gerhardt, 1994).
Mixing fluids by shaking is another source of aerosol
generation, especially larger particles. As a hot needle or
sterile loop is inserted into an agar or liquid media aero-
sols are released by the associated spattering. Similarly,
sterilizing an inoculating loop also has the potential to
create an aerosol by the same means. Inoculating rough
agar with a loop or needle also results in the production
of an aerosol, which is created by the loop or needle
vibrating over the rough surface. Water from condensa-
tion in Petri dishes can also generate an aerosol, as hap-
pens when the film between the inverted dish and the
rim of the lid is broken upon opening (Gerhardt, 1994).
Centrifugation can cause foaming which has the
potential of moistening the cap or other closure. Break-
ing that moisture film upon opening will yield an aerosol
(Gerhardt, 1994). Also, the foam bubbles may burst af-
ter the cap is removed. Predictably, droplets created
when opening containers increase the risk of contami-
nating locations proximal to their creation, including the
workspace or the user’s hands (Stimpfel & Gershey,
1991). In all of these scenarios the unidirectional airflow
of the BSC is intended to serve as a primary means of
control.
There are several benefits to using a polymerase
chain reaction (PCR)-based “release and recover” meth-
od of testing a biological safety cabinet. It is known that
even a single copy of a gene (and by extension, single
bacteria harboring that gene) can be amplified to a mil-
lion copies through the use of PCR (Gerhardt, 1994). The
PCR process therefore has a low limit of detection, and
is therefore very sensitive, potentially highly accurate
and discriminatory, and relatively quick as a method to
confirm the presence or absence of a particular segment
of DNA in a sample (Alvarez et al., 1995). Also, in situ
testing can be performed with a release and recover
approach, actually testing the functionality of a specific

www.absa.org Applied Biosafety Vol. 15, No. 4, 2010 187

Articles
cabinet for performance relative to the equipment it con-
tains. Such in-use performance testing has considerable
merit when used in conjunction with the testing of basic
design, performance and specification elements for such
user devices. Finally, the method allows for a wide varie-
ty of user-specific processes, germane to actual contami-
nation or process QA/QC issues at-hand.
In this study genetically altered E. coli colonies re-
covered from Kanamycin containing agar impaction
plates were analyzed via PCR to confirm that the colony
forming units (CFUs) collected originated solely at the
point of aerosol release, for a given set of experimental
conditions tested. Examined in this study were: 1) the
amount of cross-contamination occurring within the cabi-
net, and 2) the loss of containment from the cabinet to
surrounding workspace, potentially contaminating other
samples and equipment.
Methods
A Nuaire model NU-425-400 (Nuaire, Plymouth, MN)
Class II, Type A Biological Safety Cabinet was used
throughout this project. Annual certification of this type
of cabinet is required and was accomplished prior to the
sampling protocol (U.S. Department of Health and Hu-
man Services, 2007a, 2007b). In addition to airflow as-
sessment, filter integrity was checked with an aerosol of
poly $-olefin (PAO, LCS Inc., 2003). This aerosol was
pumped directly into the front intake manifold, and a
photometer used to scan the top diffuser, directly below
the HEPA supply filter, as well as the HEPA exhaust filter.
It should be pointed out that the introduction of PAO was
not accomplished via a “T” connection from the supply
line, and that the diffuser was not removed by the certifi-
cation contractor. Both of these deviations from NSF 49
practices are not believed to have affected the final cer-
tification determination for the cabinet, however.
Release Aerosol
To enable the specificity of the new method, a novel
test organism for release in a challenge aerosol was
needed. Novel bacterial plasmids previously created
and inserted into E. coli were utilized for this purpose
(Coschigano & Bishop, 2004). Briefly, the pCR-Blunt
II-TOPO plasmid is part of the Zero Blunt TOPO PCR
Cloning Kit (Invitrogen, Carlsbad, CA). This kit allows for
high efficiency direct insertion PCR products that have
blunt ends (Invitrogen, 2004). At each end of the open
plasmid there is a topoisomerase I enzyme that is able
to ligate double stranded (ds) DNA into the plasmid
(Shuman, 1991). In this way a circular dsDNA plasmid
can be opened and the ends activated, to ligate to an-
other strand of dsDNA (Shuman, 1994). Within the plas-
mid are several genes that allow for selection of only
those bacteria that contain the insert (Bernard et al.,
1994). Two inserts were used in this study, from the tutE
188 www.absa.org Applied Biosafety
and tutH genes of the tutE tutFDGH gene cluster in
Thauera aromatic Strain T1. These genes have been well
studied and cataloged (Coschigano & Bishop, 2004).
E. coli suspensions containing the plasmid with the
tutE insert were used in all challenge releases as well as
for positive controls. Those with the tutH insert were
used for negative controls. Cells containing the plasmid
and inserts therefore exhibited two important character-
istics making them well suited for use as challenge aero-
sol organisms. Firstly, the plasmid conferred a unique
and easily assayed antibiotic (Kanamycin) resistance.
Secondly, the plasmid allowed for a secondary confirma-
tion of the recovered test bacteria through PCR. The tutE
and tutH fragments differ in size from each other by over
500 base pairs (bp). This difference in size is easily re-
solved in agarose gels. Demonstrating the actual pres-
ence of plasmid/inserts was required to confirm that the
source of the colonies recovered was exclusively from
the aerosol released, and not from ambient sources,
preexisting cabinet contamination, or poor laboratory
techniques.
Existing strains of the tutE and tutH transformed E.
coli bacteria were used (Coschigano & Bishop, 2004).
Bacterial cultures were passed to new plates at two
week intervals to keep cells viable and maintain cell
lines. Cultures for aerosolization as well as bacterial
counting controls were grown in Luria-Bertani (LB) broth
suspensions. Millers LB Broth Base (Life Technologies,
Paisley, Scotland) was used to grow cells for eventual
aerosol preparation, and as growth media in the agar
impactor plates. LB media was made by mixing 25 g of
dry LB powder into 1 L of deionized and filter sterilized
water (dH2O), then autoclaved at 121°C for 20 minutes.
After cooling to about 40°C the media was supplement-
ed with Kanamycin (Boehringer, Mannheim, GmbH, Ger-
many). To generate the broth culture for aerosolization a
single viable colony was picked from an agar plate of
stock bacteria colonies and placed in a sterile 15 ml test
tube containing 3 ml LB broth. This was placed in a
shaking incubator at 230 RPM and 37°C and left over-
night. The following day the culture was removed from
the incubator, the suspension brought to 25 ml total
volume with LB Broth Base, and returned to the shaking
incubator for about 3 hours prior to use for creating the
challenge aerosol.
To initially determine the optimal time to grow the
25 ml cultures for aerosolization, incubator samples
were taken from 0 minutes to 4 hours. Each sample
was serially diluted and spread onto plates for enumera-
tion. An aliquot was read for optical density (OD) at 550
nm using a Carey 50 Probe UV-Visible Spectrophotome-
ter (Varian Inc., Palo Alto, CA). The use of OD at a
given wavelength for the determination of bacterial con-
centration is standard microbiological practice (Hu et
al., 2000). OD readings, in conjunction with serial plate
counts, provided a known growth time necessary for the
Vol. 15, No. 4, 2010

LB broth culture to provide a sufficient concentration of
bacteria for the challenge testing. To generate a robust
challenge culture it was found that the cultures needed
to grow for at least 3 hours in the shaking incubator at
37°C (data not shown).
The bacterial suspension concentration (E. coli/ml
of suspension) was determined for each set of release/
recovery runs via serial dilution. From those data it was
calculated that there was an average of 2.13 x 1010 E.
coli/ml of nebulizer stock (range: 1.40 x 107 to 9.50 x
1010). Because the entire E. coli suspension was nebu-
lized in each test run, the actual number of bacteria re-
leased in the cabinet for each run was known to within
an order of magnitude. Concentrations were similar to
those of Hu et al. (2000) and Ding and Wang (1997) in
their studies involving the generation of viable bacterial
aerosols. It should be appreciated that while viable cells
and spores are known to be prone to damage via the
nebulization process, the method discussed here is also
based on the recovery of a cell fragment (i.e., the tutE
or tutH genes) and its subsequent amplification via PCR.
Accordingly, this semi-quantitative approach is less de-
pendent on specific nebulizer types and release rates
than it is on quality control related to PCR. The precise
relationship between recoverable colonies on the aero-
sol sampler relative to “no hit” holes positive for the tutE
gene was beyond the scope of this preliminary work.
After growth in 25 ml of LB broth the bacteria were
harvested for nebulization. The bacterial suspension was
moved to a centrifuge tube and spun at 4355 RCF (xg)
(6000 RPM) for 10 minutes in an Avanti-J25 centrifuge
(Beckman-Coulter, Fullerton, CA). The supernatant was
discarded and the resultant bacteria pellet was suspend-
ed in 5 ml of phosphate buffered saline (PBS). PBS al-
lows suspension of the cells in an aqueous solution that
has been pH and ionically adjusted to maintain cell in-
tegrity (Hu et al., 2000). The PBS had been filter steri-
lized with a Corning 0.20 µM Sterile Syringe Filter (VWR,
Westchester, PA). For each set of experimental runs a
sample of the bacterial suspension was serial diluted,
spread onto LB agar plates, and enumerated after 2
days of growth; the remainder was used for the release
experiments that day.
Both Ding and Wang (1997) and Ranalli et al.
(2000) have shown that E. coli can be successfully aero-
solized with a nebulizer and remain highly viable for
sampling. Recovery rates of 63-91 percent were report-
ed depending on the type of aerosol sampler employed
(Ranalli et al., 2000), implying that the Collison nebulizer
they employed was capable of producing viable aerosols
with as little as 10% loss from the stock concentration
(average of 80% [79.5%] for all methods, all samplers
tested; data not shown). Several methods of bioaerosol
generation were available, and Willeke et al. (1996)
have demonstrated that the specific method of aerosoli-
zation, as opposed to the agent being used, is not usual-
www.absa.org Applied Biosafety
Articles
ly critical with regards to the viability and collection of
the aerosol particulates. The CompAir XL Compressor
Nebulizer System, Model NE-C18 (Omron, Vernon Hills,
IL) was used for all release experiments. According to
the manufacturer, the nebulizer operates at 30 to 36
psi, releasing on average 6 liters per minute of aerosol.
The pressure and flow generated by the self-contained
nebulizer-pump were not measured although the opera-
tion of this FDA-regulated medical device was visually
verified. The nebulizer generates particles from 0.5 to 5
mm in physical diameter, a range that includes most
viable environmental contaminants of interest (nano
particles and viruses excepted). The nebulizer was acti-
vated remotely while located in the BSC, and was
cleaned with 70% ethyl alcohol between runs.
The bacterial aerosol was always generated at the
same central point of the cabinet work surface (Figure
2). The outlet of the device was approximately 15 cm
above the work surface and centered within the cabinet
both laterally and from front to back. By this placement
the aerosol was generated within the normal usage area
in the cabinet but directed toward the cabinet front. Be-
cause the nebulizer expelled the entire 1 ml of bacterial
suspension it contained in 1 minute, the bioaerosol sam-
plers were started ahead of, and stopped after, nebulizer
operation so as to encompass the entire release period.
Aerosol Recovery
An array of 3 single-stage (N-6) bioaerosol impactors
(Aerotech 6, Phoenix, AZ) was typically used for aerosol
recovery. This sampling technology was first described
by Andersen in 1958 and is employed essentially un-
changed in the samplers utilized (Andersen, 1958). In
use, each sampler pump was set to a flow rate of 28.3
liters of air per minute as indicated on the integral
rotameter/flow controller, providing a sampling rate
equivalent to the 1 cubic foot of air per minute design
specification for the original Andersen sampler. Because
the pumps for the samplers were located in the same
room as the samplers, aerosols not trapped by the im-
pactor could potentially be vented directly into the room,
thereby resulting in erroneously high results. To prevent
such contamination 0.2 µm Pall filters (VWR, West Ches-
ter, PA) were attached to each pump exhaust.
Sterile media plates used in the bioaerosol samplers
were prepared in the authors’ laboratory. Each plate
consisted of a 100 mm x 15 mm sterile polystyrene Petri
dish (Fisher Scientific, Pittsburgh, PA) containing exactly
36 ml of LB agar. That volume of agar resulted in the
appropriate agar height required by the design of the
sampler for the specified collection efficiency of particles
in the N-6 size range (Andersen, 1958). Prior to and af-
ter each run the cabinet and surfaces were thoroughly
cleaned. Surfaces were sprayed with 70% ethanol and
the cabinet was allowed to run for % 1 hour between
runs with the UV lamp on. After the UV exposure period
Vol. 15, No. 4, 2010 189
Articles
Figure 2
Nebulizer and 3 Bioaerosol Samplers within the BSC. Note: Power outlets were remotely operated.
but prior to any aerosol release, negative controls were
collected by sampling the air in the same manner and at
the same locations as would be used for the subsequent
challenge release. The nebulizer was not running and no
bacteria were released during this sampling.
Before each experimental run, a negative control
sample (i.e., no aerosol nebulized) was collected under
the same conditions and identical locations used in
the experimental runs. All negative controls showed no
growth (n = 24), demonstrating that the methods used
to clean the cabinet between runs were effective, that
the laboratory and cabinet were free of airborne recom-
binant E. coli contamination, and that the colonies recov-
ered during the experimental runs originated solely from
the test aerosols generated. A positive control was also
run for each experimental challenge. For the positive
control runs, all plates (n = 24) grew over 300 colonies
of the Kanamycin-resistant bacteria (i.e., colonies were
found under each of the sampler’s 300 holes). As viable
bacteria are drawn though the holes multiple CFUs going
through the same hole will cause colony masking, such
that 1 colony recovered may in fact have been the result
of several impacting the plate beneath a given hole. Pos-
itive hole correction is therefore necessary, as detailed
by Andersen (1958). A positive hole count of 300 colo-
nies corresponds to a corrected particle count of 555+
colonies, corresponding to an aerosol concentration of
> 9.91 x 103 E. coli/m3. At these high numbers, the cor-
rected count is therefore a minimum estimate of the
total colony forming units present.
Samplers were placed at 1 of 4 locations for release
and recovery experiments. The first location, used as a
control for all runs, was directly in front of and under the
aerosol plume generated from the nebulizer. The second
and third locations were to the left and right of the cen-
ter work space in the cabinet, 30 cm from the center
190 www.absa.org Applied Biosafety
where the aerosol was generated, and equidistant from
the front grill and back wall. This distance was within the
normal workspace of the cabinet but not close enough
to the walls so as to be out of the area normally used by
cabinet operators. The fourth location was outside the
cabinet, on a stand in front of the front opening of the
cabinet on the center line, directly in front of the nebuliz-
er, 20 cm from the front sash of the cabinet and 5 cm
below the work surface of the cabinet.
After aerosol recovery, plates were incubated for 2
days and results determined. Concentrations were calcu-
lated by counting the colonies found on the sample
plates, applying the positive hole correction, dividing the
total volume of air sampled and averaging for all repli-
cates (Equation 1). After counting colonies, 3 colonies on
each plate were randomly selected for definitive identifi-
cation by PCR.
(#CFUs [after positive hole correction])
= CFUs/m3 (1)
(L/min[flow rate])(sampling time min)(1 m3/1000L)
PCR Quantitation
The primers used for all experimental and control
runs were the M13 Forward (TOP-MR) and M13 Reverse
(TOP-MF) primers (IDT, Coralville, IA). The TOP-MR primer
sequence is 5'-CAG GAA ACA GCT ATG AC and the TOP-
MF primer sequence is 5'-GTA AAA CGA CCAA C. These
were supplied as dried powders and dissolved in PCR
grade dH2O (PCR-dH2O) (Qiagen Scientific, Valencia, CA)
to a concentration of 2 µM. The 10 X reaction buffer
used was ThermoPol Buffer (New England Biolabs, Ips-
wich, MA) and was supplied with the Taq polymerase
(Taq DNA polymerase is the enzyme responsible for rep-
licating the DNA). The stock dNTP solution contained 10
mM of each of the four dNTPs: dATP, dCTP, dGTP, and
Vol. 15, No. 4, 2010
 Articles

dTTP (Invitrogen, Carlsbad, CA). All stock and intermedi-
ate dilutions of PCR reagents were kept at -20°C be-
tween uses. To avoid fractionalization, solutions were
thawed at room temperature out of direct sunlight and
refrozen by placing directly into the -20°C freezer. Posi-
tive controls confirmed the functionality of all reagents.
The time and temperature of each PCR step was the
same for each experimental run and every control run.
All PCR runs were conducted in a PTC-100 Programma-
ble Thermal Controller (MJ Research Inc., Watertown,
MA). Briefly, the first step in the reaction was for 2
minutes at 95°C, to lyse the bacteria and melt apart
dsDNA, forming the ssDNA necessary to allow subse-
quent binding of primers and enzymes. The second step,
significant only after the first full cycle (e.g., following
Step 5), was 30 seconds at 95°C. The third step was
for 30 seconds at 49°C, the optimal temperature at
which these primers bind. In the fourth step (1 minute
at 72°C), primers were extended. The fifth step was a
return to 95°C for 30 seconds, or Step 2. The cycle was
repeated 29 more times for a total of 30 cycles. After
the thirtieth cycling the reaction progressed to Step 6,
a final primer extension step at 72°C. Step 6 was run for
7 minutes to ensure all primers (and any strands only
partially extended in previous cycles) were completely
extended. At the conclusion of Step 6 the PCR machine
cooled the reactions to 4°C, concluding the amplifica-
tion process.
As an internal check, PCR was run on the TOPO plas-
mid containing the previously ligated tutE insert. The
fragment produced from that PCR run was 447 bp long.
PCR was run both on this plasmid in solution, as a con-
trol, and in viable bacteria. These controls are found as
Lane 1 and Lane 4, respectively, in Figure 3. The band in
Lane 1 is slightly smaller than the 500 bp band seen in
the 100 bp ladder lane, Lane 2. This is consistent with
the expected size of a PCR amplified segment of 447 bp.
The difference in fluorescent intensity between Lane 1
and Lane 4 is expected and not uncommon. Lane 3, a
negative control, was identical to other lanes except that
no DNA was added to the reagents for PCR. The frag-
ment produced from PCR run on the bacteria containing
the tutH insert in the TOPO plasmid was expected to be
1093 bp long. The size was confirmed by PCR and can
be found in Lane 5 on the gel in Figure 3.
After PCR the amplified gene sample was run on a
gel for viewing and confirmation. An appropriate volume
of concentrated loading dye (Promega, Madison, WI) was
added to each PCR tube. The PCR product was then
loaded into wells in a 1%/1%, wt/wt, Metaphore Agarose
(Bio Whittaker Molecular Applications, Rockland, ME)/
Agarose gel (Low Electroendosmosis Agarose, National
Diagnostics, Atlanta, GA). Per standard protocols 1% wt/
vol Metaphor Agarose and 1% wt/vol Agarose gel mix
was made in 200 ml batches, to which 10 ml 1% ethidi-
um bromide (Fisher Scientific, Pittsburgh, PA) was add-
ed. Each PCR reaction product was 20 µl, and after the
reactions the volume was adjusted to 24 µl with loading
dye. Of that 24 µl, 8 µl were typically loaded into a well
for electrophoresis. Gels were allowed to run at 60 volts

Figure 3
PCR Controls
Lane 1 is the positive control from the plasmid and tutE insert. Lane 2 is the base pair size marker; smaller fragments
move more quickly through the gel and are found lower. Lane 3 is a negative gel control (no DNA). The faint band in
Lane 4 indicates the typical limit of detection with PCR, and shows the bacterial positive control. Lane 5 is a negative
control from bacteria with only the larger tutH insert. Lane 6 is from a recovered bacterial bioaerosol colony.

www.absa.org Applied Biosafety Vol. 15, No. 4, 2010 191

Articles

for approximately 90 minutes in running buffer using
a Power Pac 300 power supply (Bio-Rad Laboratories,
Hercules, CA). Gels were viewed in a Bio-Rad Universal
Hood with live video feed (Bio-Rad Laboratories, Segrate,
Milan, Italy) using the Bio-Rad “Quantity One” software
package (Bio-Rad Laboratories, Hercules, CA). Images
were captured with the manual exposure setting adjust-
ed for maximum separation of bands from background.
Several internal PCR controls were included in the
electrophoresis gels to confirm that the experimental
bands produced were from the tutE insert. Those PCR
controls included plasmid DNA in solution, negative and
positive control bacteria as well as a blank (PCR reagent
only) negative control lane (Figure 3). For each gel run, a
negative and a positive bacterial control was also includ-
ed. Those controls were consistently found in nine repli-
cates. PCR was run on the purified plasmid containing
the tutE insert. The expected band is found in Lane 1.
Lane 3 is the result of the first negative control experi-
ment. Lane 4 is a positive control from the original bac-
terial stock. Lane 5 is from PCR run on bacteria trans-
formed with the tutH insert. (This lane functioned as
both a negative and positive control, since this strain
contains no tutE DNA and therefore should have shown
no band at or near 447 bp.) Lane 5 functioned as a posi-
tive control for the primers, TOP-MR and TOP-MF. The
PCR product in Lane 5 was expected to be slightly above
the 1000 bp marker band, and those expectations were
confirmed. The size differences of these bands allow for
excellent discrimination. Lane 6 illustrates a positive
control from one of the aerosol release experiments.
After the E. coli/m3 concentrations were quantified
from the aerosols impacted onto the plates, PCR was
performed confirming that recovered colonies resulted
exclusively from the test aerosol released and not from
other sources. A representative gel showing experi-
mental recovery runs, Lanes 6-13, is reproduced in Fig-
ure 4. Randomly selected colonies from the bioaerosol
recovery plates from each sampling location were sub-
jected to PCR evaluation. Up to 3 colonies, when pre-
sent, were tested from each plate. PCR confirmed that
the plasmid in the original bacteria released in the cabi-
net was the same as that found on the bioaerosol
plates. For the positive control plates, where the number
of colonies was always > 300, this resulted in a sam-
pling rate of only 1%. Such sampling was deemed appro-
priate given the consistent appearance of colonies on
such plates.
Results
Limits of Detection
Since the nebulizer generates ~6 L of aerosol per
minute, effectively turning 1 ml of the bacterial chal-
lenge agent into 6 L of air containing that suspension in
particulate form the challenge aerosol was easily esti-
mated. The average concentration of bacteria for each
challenge was 1.70 x 1010 E. coli/ml (as determined by
OD and serial dilution plate counting), or 2.84 x 109

Figure 4
Representative Electrophoresis Gel
Lane 1 is a positive control from DNA in solution. Lane 2 is the base pair size marker. Lane 3 is a DNA-free
(negative) control. Lanes 4 and 5 are a positive and negative control, respectively, from stock bacteria.
Lanes 6 through 11 are from colonies recovered directly below the nebulizer. Lanes 12 and 13 are
from separate colonies recovered from a bioaerosol sampler outside the hood (blower off).

192 www.absa.org Applied Biosafety Vol. 15, No. 4, 2010


E. coli/L, and 2.84 x 1012 E. coli/m3 in aerosol form. The-
se values include an estimated 20% reduction in E. coli
collection efficiency based on the data of Rinalli et al.
(2000). This downward correction was applied to ac-
count for decreased E. coli viability resulting from the
nebulization process, as well as drying effects related to
evaporation throughout the process.
Colonies counted during aerosol recovery can be
expressed as E. coli per cubic meter (m3) of air. Equation
1 was used to determine this concentration. Terms in
Equation 1 include the “#CFUs,” the colony count from
each plate for a given run adjusted by the positive hole
correction. The “L/min [flow rate]” was 28 L/min and
“(sampling time-min)” was the sampling time (2 minutes
for all runs). The final answer, in CFU/m3, describes the
number of E. coli found in 1 cubic meter of air. The total
CFU counts from the center sampling location during
cabinet operation were calculated using Equation 1, af-
ter application of the positive hole correction (Table 1.)
With the cabinet off, the concentrations recovered
averaged > 9.91 x 103 E. coli/m3, but with the cabinet
operating only 0.89 E. coli/m3 were recovered on aver-
age (3 trials). This value represented the lower limit of
detection as empirically determined. The upper limit of
detection in this study was therefore 9.91 x 103 E. coli/
m3 which represented aerosol sampling plates of > 300
colonies (and > 555 CFUs based on the positive hole
correction). The concentration recovered divided by the
concentration released results in the method sensitivity
of 2.52 x 10-13 E. coli/m3. These data demonstrate in an
applied sense that an operational, properly functioning
BSC results in an actual aerosol reduction of approxi-
mately 13 orders of magnitude (1012 E. Coli/m3 ! 0.89
E. coli/m3).
For every release experiment a positive control sam-
ple plate located directly under the point of aerosol gen-
eration was collected. As expected, those plates all
showed the maximal growth of greater than 300 colo-
nies. From each of these positive recovery plates 3 colo-
nies (1%) were subjected to PCR analysis. Lane 6 on
Figure 3 is the result of that PCR on the first positive
control plate. Notice that the size of band produced from
that first control colony is the same size in bp as the
positive controls on that gel, Lanes 1 and 4. The positive
result from Lane 3 demonstrated that the bacteria could
successfully be aerosolized, collected and confirmed
as the original test bacteria through PCR. On each gel
(n = 8) both a negative and positive control were run, as
well as the base pair markers.
Since N-6 bioaerosol plate media were inhibited with
Kanamycin, and negative controls ruled out the pres-
ence of other bacteria, any colony growth is conclusively
known to be the result of the genetically modified bacte-
ria alone. PCR was run to confirm these expectations,
and each run condition and location was accompanied
by at least 3 such control replicates.
www.absa.org Applied Biosafety
Articles
Cross-contamination
As anticipated, internal cross-contamination was
largely a function of whether the cabinet was on or off.
The center location consistently showed 300+ colonies,
or greater than 555 CFUs after positive hole correction,
with the cabinet either on or off. Samplers set up 30 cm
to the left and right of the center of the workspace
showed results entirely dependent on the cabinet being
on or off. With the cabinet off the lateral locations pro-
duced greater than 555 E. coli per plate for every run.
With the cabinet operating the E. coli count was dramati-
cally lower with side locations producing only a total of 3
E. coli colonies in all 3 replicate runs.
The extent of lateral bioaerosol migration within the
cabinet was less than a single bacterium (0.89 E. coli/
m3) as reported in the average of the E. coli/m3 from
both the right and left samplers.
Containment
Sampling outside of the cabinet was done to exam-
ine and quantitate the escape of viable aerosol. Alt-
hough there is no expectation a Class II, Type A BSC is
100% effective for containment, results demonstrated
zero (0) E. coli released while the cabinet was running.
This was not unexpected in that the NSF method pur-
posely introduces an airflow disruption in order to pro-
duce non-zero results. Nevertheless, the PCR technique
allowed an empirical observation of the actual degree of
such control under in situ conditions. All plates used for
sampling outside of the cabinet either had no colonies
(cabinet on), or greater than 300 (cabinet off). Figure 5,
Lanes 2-3 and 18-21 show the results of PCR on the
colonies selected from samplers outside the cabinet
(cabinet off). These results demonstrate the ability to
recover bacteria that were aerosolized inside the cabinet
once they have migrated to the cabinet exterior, as well
as show good experimental hygiene with respect to the
aerosolized DNA from previous experimental runs.
The positive and negative bioaerosol plate controls
were also confirmed with PCR. All negative controls had
no (zero) growth, while each positive control plate had
over 300 colonies. All colonies selected for PCR showed
a positive band for the expected genetic insert. A posi-
tive control plate was run for each experimental condi-
tion. A selection of those PCR results for both the cabi-
net on and the cabinet off conditions are found in Lanes
6-11 in Figure 3, and Lanes 4-7, 14, 18-22 in Figure 5.
All PCR results showed a positive band except for
the negative controls. Including the control runs, there
were 75 PCR runs conducted to verify bioaerosol colony
identities. Twenty of the 75 runs are presented in Figure
5, shown on 2 gels runs simultaneously. PCR confirmed
that the collected bacteria under all conditions and in all
locations were in fact the recombinant E. coli released
for each particular run.
Vol. 15, No. 4, 2010 193
Articles

Discussion developing this new assessment technique. Since in-

From these results it is clear that a properly operat-
ing BSC does an excellent job of containing even ex-
tremely concentrated aerosols generated within. There
were no bacteria recovered outside of the cabinet when
the cabinet was running under normal, undisturbed op-
erating conditions. The effect of operator movements,
procedures, and equipment were not examined while
ward airflow is critical to cabinet containment, and any
operator effects would disturb that flow, releases under
such conditions would be expected. With the sensitivity
of this new method it is evident that any such releases
related to operator effects could be detected. The cross-
contamination findings, demonstrating detectable lateral
migration even while the cabinet was on, affirm this as-
sertion.

Figure 5
PCR Results from 20 Experimental Colonies
Note the consistent and strong bands at ~500 bp location of the positive controls. Lanes 2, 3, and 18-21 are all from
recovered bioaerosol colonies from outside the hood. Lanes 4-7 are from the center location with the hood on. Lanes
8 and 13 are from the left location and Lanes 9-12, 16, and 17 are from the right location, all within the hood. Lanes
14 and 18-22 are from the center location. Lanes 23 and 24 are positive and negative bacterial controls, respectively.

Table 1
Bacterial Concentrations Recovered by Hood Location. Each value reported
is the average three replicates, which accounts for non-integer values.

Sampler Locations Hood On Hood Off

Within the Hood E. coli/m3 E. coli/m3
Center > 9.91 x 103 > 9.91 x 103
Left of Center (30 cm) 1.19 > 9.91 x 103
Right of Center (30 cm) 0.60 > 9.91 x 103

Sampler Location Hood On Hood Off

Outside of the Hood E. coli/m3 E. coli/m3
Center (only) < 0.0 > 9.91 x 103

194 www.absa.org Applied Biosafety Vol. 15, No. 4, 2010


PCR was found to be an excellent addition to simple
bioaerosol counts for cabinet testing. The use of PCR with
known primers and inserts, in conjunction with appropri-
ate experimental controls, provided an excellent tool to
demonstrate the movement of bacteria laterally within a
BSC. Results demonstrated a cross-contamination poten-
tial of 0.6-1.2 E. coli/m3 at the 30 cm distances exam-
ined (Table 1). The percent reduction can be defined as
the concentration of the bioaerosol sampled during the
recovery sampling divided by the concentration collected
during the positive control runs. Using the concentra-
tions determined while the cabinet was off as the base
line, the amount of lateral aerosol migration to both the
right and left samplers showed an average of 9.91 x 103
E. coli/m3 (3 replicates). While the cabinet was on identi-
cal sampling demonstrated an average of < 1 E. coli/m3
detected (8.93 x 10-1 E. coli/m3). This concentration re-
duction was almost 5 orders of magnitude (e.g., an aver-
age 9.01 x 10-3 percent reduction), which roughly agrees
with the minimum 4 orders of magnitude reduction re-
quired in NSF 49. This percentage drop more accurately
demonstrates the cabinet’s ability to limit cross-
contamination under these conditions, and represents a
new metric of interest to many BSC users. Furthermore,
this specific test is markedly more meaningful than the
assessment of unidirectional flow performed during nor-
mal NSF 49 style field testing. Using PCR to confirm that
the bacteria recovered were the same as released, in
conjunction with the simultaneous use of conventional
air sampling techniques to enumerate that movement,
was consistent with the expectation that internal cabinet
contamination is likely even under the best (i.e., undis-
turbed) of circumstances. With the added problems from
operator effects, such cross-contamination presents the
greatest overall risk both internally and externally
(through transferable operator or equipment contamina-
tion). Aerosol containment under usual BSC use condi-
tions and procedural manipulations was not examined.
Nor was the performance of the BSC after certain higher
risk release events, such as filter changes, blower re-
pairs, spills, power loss, or cabinet relocation. Such con-
siderations are highly relevant to future studies.
This work introduces the possibility of testing kits for
in situ BSC evaluations by suitably qualified end users.
Many equipment requirements for the utilization of such
a kit typically exist in many BSC-equipped laboratories,
and could easily be applied to this in situ testing. In this
regard end users would only need to purchase or rent
the missing components. In this way end users would be
able to assay their BSCs as their own schedule allowed
and in a way that had increased accuracy and sensitivity
over current field tests. Perhaps of most significance,
focused studies to determine recalcitrant contamination
problems could also be conducted to improve produc-
tion, quality control or both.
Data demonstrate that the BSC can produce an aer-
www.absa.org Applied Biosafety
Articles
osol reduction of up to 13 orders of magnitude in the
lateral motion of a particulate under ideal conditions.
This is much higher than the NSF microbial aerosol chal-
lenge method which produces 8 orders of magnitude
difference for the personnel protection test and 4 orders
for the cross-contamination test. Possible explanations
for this higher sensitivity include our use of a different
test procedure involving different release and collection
locations, and our testing under ideal circumstances,
including closed room door and no intentional disruption
of the inward room airflow to the cabinet (Figure 1 “A”).
While this sensitivity may in fact not be required for rou-
tine manipulations or agents of lower biosafety level (1),
its utility might best be realized in the study and assign-
ment of biosafety levels to new agents or new equip-
ment. Such low limits of detectability as demonstrated
here (~1 E. coli/m3) could also be very useful in training
new BSC operators in good technique and practices.
Furthermore, an entirely new area of research is now
available given the sensitivity of the novel technique.
Quantitative risk assessments are now readily and safely
possible to better determine and describe release poten-
tials related to traditional microbiological applications as
well as new methods or equipment.
Conclusions
The primary aim of this study was to explore the fea-
sibility of a more discriminatory test of the BSC (relative
to the field tests specified in NSF 49 Annex F). This was
accomplished by combining traditional bioaerosol sam-
pling with the sensitivity and accuracy permitted by PCR
on specific recombinant bacterial genes. This tested the
extent to which a biological safety cabinet was able to
control and contain a true bioaerosol under in situ condi-
tions (as it is primarily designed to do). The creation of a
novel experimental test for BSCs was also achieved, in
which was demonstrated the effectiveness of using PCR
in conjunction with conventional air sampling methods.
In conducting this research it was shown that:
a. A properly functioning biological safety cabinet was
able to control the egress of a bioaerosol to within 13
orders of magnitude under steady-state conditions.
b. The reduction of lateral motion of an E. coli aerosol
by a Class II, Type A BSC is approximately 5 orders of
magnitude, a result confirming this ability of BSCs. The
BSC tested was unable to fully prevent lateral motion of
a particulate under the conditions tested.
c. PCR can be used to compliment conventional NSF
49 field testing techniques with an aerosolized challenge
of viable bacteria in order to create a quantitative assay
and useful in situ test method for cabinet end users.
The applicability of such work is not restricted to the
microbiological or biotechnology communities. Others
who work with or must contend with small particle con-
taminants include microelectronics and nanotechnology
Vol. 15, No. 4, 2010 195
 Articles
manufacturers, and all users of ISO (or older FS209)
classified cleanrooms. In such settings this technique
could be used, or adapted, for the conduct of additional
studies using the E. coli aerosol as a surrogate for the
actual particle of interest.
Future studies of this novel technique might exam-
ine the abilities of BSCs to prevent the ingress of particu-
lates at the front portal when other than undisturbed,
steady-state airflow conditions exist. Furthermore, it
would be of great interest to further examine lateral mi-
gration of bioaerosols given users’ arm movements. For
example, a linear array of samplers abutting each other,
moving away from the nebulizer and parallel to the front
portal, could be used to better quantitate actual cross-
contamination potential. Also of interest is the ability of
the BSC to contain and control bioaerosols in the pres-
ence of routine laboratory cabinet use. Testing using the
novel method might be repeated when equipment com-
monly used and stored within a BSC is present, such as
pipettes, Bunsen burners, reagent bottles, etc. Addition-
ally, the effect of the operator or an operator surrogate
(i.e., a mannequin), could be used to examine the ef-
fects of arms changing the air patterns both at the front
intake and within the cabinet, in conjunction with such
operations. Finally, direct comparison of this method to
the NSF 49 microbiological aerosol challenge test
should be conducted in future work on its development
or validation.
Disclaimer
Financial Considerations: none
References
Alvarez, A. J., Buttner, M. P., & Stetzenbach, L. D. (1995). PCR
for bioaerosol monitoring: Sensitivity and environmental
interference. Applied and Environmental Microbiology, 61,
3639-3644.
Andersen, A. (1958). New sampler for the collection, sizing,
and enumeration of viable airborne particles. Journal of Bac-
teriology, 76, 471-484.
Bernard, P., Gabant, P., Bahassi, E. M., & Couturier, M. (1994).
Positive selection vectors using the F plasmid ccdB killer
gene. Journal of Gene Medicine (Gene), 148, 71-74.
196 www.absa.org Applied Biosafety
View publication stats
Coschigano, P., & Bishop, B. (2004). Role of benzylsucinate in
the induction of the tutE tutFDGH gene complex of T. aro-
matica strain T1. FEMS Microbiology Letters, 231, 261-266.
Ding, P. H., & Wang, C. S. (1997). Effect of sampling time on
the sampling efficiency of all-glass impinger-30 samplers for
E-coli. Aerosol. Journal of Aerosol Science, 28, S671-S672.
Gerhardt, P. (Ed.). (1994). Methods for General and Molecular
Bacteriology. Washington, DC: American Society for Microbi-
ology (ASM).
Heinsohn, P., Jacobs, R., & Concoby, B. (1995). Biosafety Ref-
erence Manual, (2nd ed.). American Industrial Hygiene Asso-
ciation Biosafety Committee. Fairfax, VA: AIHA Publications.
Hu, W-G., Chen, J., Collins, F. M., & Gu, X-X. (2000). An aerosol
challenge mouse model for Moraxella catarrhalis. Vaccine,
18, 799-804.
Invitrogen. (2004). Zero Blunt TOPO PCR Cloning Kit, Version L.
Invitrogen Instruction Manual. 25-0215. Carlsbad, CA.
LCS, Inc. (2003). Laminar Flow Device Certification Report.
Laboratory Certification Services, Inc., Columbus, OH.
NSF International. (2008). NSF/ANSI 49 - 2009 Biosafety Cabi-
netry: Design, Construction, Performance, and Field Certifica-
tion. [NSF/ANSI Standard 49-2008]. Ann Arbor, MI: NSF.
Ranalli, G., Principi, P., & Sorlini, C. (2000). Bacterial aerosol
emission from wastewater treatment plants: Culture meth-
ods and bio-molecular tools. Aerobiologia, 16, 39-46.
Shuman, S. (1991). Recombination mediated by vaccinia virus
DNA topoisomerase I in Escherichia coli is sequence specif-
ic. Proceedings of the National Academy of Sciences, 88,
10104-10108.
Shuman, S. (1994). Novel approach to molecular cloning and
polynucleotide synthesis using vaccinia DNA topoisomerase.
Journal of Biological Chemistry, 269, 32678-32584.
Stimpfel, T. M., & Gershey, E. L. (1991). Design modifications
of a Class II Biological Safety Cabinet and user guidelines for
enhancing safety. American Industrial Hygiene Association
Journal, 52(1), 1-5.
U.S. Department of Health and Human Services, Centers for
Disease Control and Prevention, & National Institutes of
Health. (2007a). Biosafety in Microbiological and Biomedical
Laboratories (5th ed.). L. C. Chosewood & D. E. Wilson (Eds.).
Washington, DC: U.S. Government Printing Office.
U.S. Department of Health and Human Services, Centers for
Disease Control and Prevention. (2007b). Primary Contain-
ment for Biohazards: Selection, Installation and Use of Bio-
logical Safety Cabinets. L. C. Chosewood & D. E. Wilson
(Eds.). Atlanta, GA.
Willeke, K., Ulevicius, V., Reponen, T., & Grinshpun, S. A. (1996).
Development and evaluation of aerosol generators for biologi-
cal materials. Journal of Aerosol Science, 27, S249-S250.
Vol. 15, No. 4, 2010